Jackson Loth

Profile

Project title:

Time to vibe together: cloud-based guitar and intelligent agent

Abstract:

This PhD will investigate how to augment a traditional acoustic guitar using artificial intelligence and a cloud-based environment supporting music production and content sharing. A cloud-based intelligent agent aimed at recommending or generating content for guitar players will be researched, developed and assessed. The methodology will rely on automatic analyses of musical expression from signals produced on the guitar and the development of web-based interfaces for user interaction. Machine learning techniques such as deep learning will be used to train a content recommender or generator system from audio data (e.g. guitar chord progressions, melodies and riffs) and semantic information describing the musical context (e.g. chord harmonic function, musical style). Graph-based approaches will be researched taking into account user preferences, the musical compatibility of audio items based on audio engineering considerations, and music theory. The intelligent agent will be assessed through user evaluations using human-computer interaction methods for music making and creativity support. The PhD will also research how the technology could be applied to foster an online community of smart guitar users. Examples of applications include recommendation of accompaniment tracks, sound fx, or content for music learning; automatic harmonisation of melodies; control of guitar tone to adjust audio effects or simulate a guitar amp based on a reference

C4DM theme affiliation:

Augmented Instruments, Sound Synthesis

Research